remain. What is perhaps more important is the fact that the effects of family size are evident within each class, even the highest. Direct Evidence There is one study, to my knowledge, in which an hypothesis concerning the nature of the effect of family size was developed and tested. Nisbet tested his hypothesis concerning effects on intelligence among Aberdeen school children (74). His subjects were the children passing from primary to secondary school in 1949 and 1959—around 2,500 each year—who were given a battery of intelligence tests. His idea was that the greater adult contact to be expected in smaller families would stimulate the development of better verbal ability and that this would account for some of the association between family size and intelligence test performance. He tested this first by examining the association between verbal score (English attainment) and family size, while holding intelligence scores constant. He found, "All these partial correlations are negative and significantly different from zero" (i.e., highly unlikely to occur because of sampling fluctuations). Second, he compared the negative correlation with family size in tests dependent on verbal ability with those of tests more or less independent of such ability. He expected, and found, a greater degree of negative correlation in the verbal test, though the difference was not too great. Finally, he examined the negative correlation at various ages on the assumption that the effect of the environmental influence—the decreased contact with adults in larger families—"will tend to be greater at later ages when the cumulative effect of the environment begins to show itself." The negative correlations were indeed found to increase, both when different groups of children, aged 7, 9, and 11, were compared* and also when the results at ages 7, 9, and 11 of the same children were compared.' Nisbet concludes . . . that part of (though not all) the negative correlation of family size and intelligence test score may be attributed to an environmental influence of the size of family on verbal development and through it on general mental development. (74, p. 286) Here, then, appears to be some clarification of the role of one element, contact with adults, which may account for variation in intelligence with family size. His data also show that the effect increases with increasing age of the child. Whatever the precise causal interrelations may be, it is clear that the process continues throughout childhood. Vs = -.256,-.287, and-.333 respectively. Vs = -.209, -.226, and -.289 respectively.pment, providing the pressure of numbers is not too great. On the other hand, family size has no effect in the higher social classes.